Judgment and Decision Making, Vol. 6, No. 8, December 2011, pp. 771–781
Measuring Social Value Orientation
Ryan O. Murphy∗
Kurt A. Ackermann†
Michel J. J. Handgraaf‡
Abstract
Narrow self-interest is often used as a simplifying assumption when studying people making decisions in social contexts.
Nonetheless, people exhibit a wide range of different motivations when choosing unilaterally among interdependent
outcomes. Measuring the magnitude of the concern people have for others, sometimes called Social Value Orientation
(SVO), has been an interest of many social scientists for decades and several different measurement methods have been
developed so far. Here we introduce a new measure of SVO that has several advantages over existent methods. A
detailed description of the new measurement method is presented, along with norming data that provide evidence of its
solid psychometric properties. We conclude with a brief discussion of the research streams that would benefit from a
more sensitive and higher resolution measure of SVO, and extend an invitation to others to use this new measure which
is freely available.
Keywords: Social Value Orientation (SVO), social preferences, narrow self-interest, measurement methods, individual
differences.
1 Introduction
The assumption of narrow self-interest is central to rational choice theory. The postulate is that decision makers
(DMs) are concerned about maximizing their own material gain, indifferent to the payoffs of other DMs around
them. This is a simplifying assumption that yields a
powerful framework to predict and explain human decision making behavior across a wide variety of domains.
However there are reliable counterexamples demonstrating that DMs’ elicited preferences and choices are often influenced in part by the payoffs of other DMs, thus
challenging what some have termed the selfishness axiom
(Henrich et al., 2005).
Studies on the motivations that underlie interdependent
decision behavior have a long history and these motivations have been referred to by a variety of names, including: social preferences, social motives, other-regarding
preferences, welfare tradeoff ratios, and Social Value Orientation (SVO). For consistency, we refer to this construct as SVO for the remainder of this paper. Within
the SVO framework it is assumed that people vary in
their motivations or goals when evaluating different reThis research has been supported in part by United States National
Science Foundation grant SES-0637151. Additional thanks to Mike
Kuhlman, Jeff Joireman, Michael Schulte-Mecklenbeck, and Miguel
Fonseca for their constructive feedback on early versions of this paper.
∗ ETH Zürich, Chair of Decision Theory and Behavioral Game
Theory, Clausiusstrasse 50, 8092 Zürich, Switzerland. Email: rmurphy@ethz.ch.
† ETH Zürich, Chair of Decision Theory and Behavioral Game Theory, Clausiusstrasse 50, 8092 Zürich, Switzerland.
‡ Wageningen University, Economics of Consumers and Households
(ECH), Hollandseweg 1, 6706 KN, Wageningen, The Netherlands.
source allocations between themselves and another person. As examples, a DM may endeavor to maximize her
own payoff (individualistic), maximize (competitive) or
minimize (inequality averse) the difference between her
own and the other person’s payoff, or maximize joint payoffs (prosocial). It is worth noting, however, that the assumption of narrow self-interest is itself a particular SVO,
namely a perfectly individualistic orientation. Moreover
considering a spectrum of different SVOs is not a challenge to rational choice theory per se, but rather the extension of a postulate in an effort to increase the theory’s
psychological realism and descriptive accuracy.
SVO has been found to affect cognitions and account
for behavior across a range of interpersonal decision making contexts, specifically in the domain of negotiation settings (De Dreu & Boles, 1998) and resource dilemmas
(Roch et al., 2000; Roch & Samuelson, 1997; Samuelson,
1993). SVO has also been identified as a covariate, interacting with different emotional states and influencing the
propensity to cooperate (Zeelenberg, Nelissen, Breugelmans & Pieters, 2008). SVOs have even been identified
in non-human primates (Burkart, Fehr, Efferson & van
Schaik, 2007), indicating that some other species also
show intrinsic preferences for prosocial behavior.
In order to use the full explanatory power of SVO as
a psychological construct, we need to measure it efficiently, reliably and validly. Several different measurement methods for quantifying variations in SVO across
individuals have been developed (for overviews, see McClintock & Van Avermaet, 1982; Au & Kwong, 2004;
Murphy & Ackermann, 2011). Although the use of existent measures has produced a wealth of findings even
771
Judgment and Decision Making, Vol. 6, No. 8, December 2011
with categorical approaches (see, for instance, De Dreu
& Boles, 1998; Kuhlman & Marshello, 1975a, 1975b;
Van Lange & Visser, 1999), these measures have substantial limitations. For instance, some measures yield
only low-resolution output that lack sensitivity to important individual differences, providing at best a nominal
categorization (e.g., the Triple-Dominance Measure, see
Van Lange, Otten, De Bruin & Joireman, 1997). Other
measures are highly inefficient and often fail to produce
consistent results for a substantial proportion of subjects
(e.g., the Ring Measure, see Liebrand, 1984). Yet other
methods require substantial time and effort from a research subject in order to produce a score (e.g., Utility and Conjoint Measurement procedures, or Regression
and Clustering techniques, see Wyer, 1969; Radzicki,
1976; or Knight & Dubro, 1984, respectively). Moreover, none of these existent measures are explicitly designed to detect more nuanced motivations like inequality
aversion. Specifically, previous measures have not disentangled the orientation of joint gain maximization from
the motivation to minimize the difference between outcomes. Although these two orientations are related in that
they both indicate a deviation from individualism towards
prosociality, they are substantially different motivations,
which should be differentiated both theoretically and operationally.
Furthermore, Social Value Orientation is a continuous
construct, as it corresponds to the quantity of how much
a DM is willing to sacrifice in order to make another DM
better off (or perhaps worse off). This quantification of
interdependent utilities can be best represented on a continuous scale. Moreover, since the most commonly used
SVO measures to date produce only categorical data, a
substantial amount of information related to peoples’ social preferences is being discarded and ignored. Consequently, the full explanatory power of SVO has not been
used because of this unnecessary sacrifice of statistical
power (see Cohen [1983] for a discussion of the unfortunate practice of reducing continuous variables to categories).
In our view, a method for assessing SVO should yield
high resolution output which makes it sensitive to interand intra-individual differences and facilitate comparisons thereof, be easy to use, be efficient, be able to detect the most prevalent SVO individual differences, allow for an evaluation of rank orders of social preferences, and yield meaningful results for virtually all subjects. Amongst these criteria, we consider the demand
for a high resolution measure which produces data on a
continuous scale as crucial.
We introduce here a new measure of SVO, which takes
this conceptualization into account and allows for greater
explanatory potential of SVO through increased statistical power while also meeting the aforementioned psycho-
772
Measuring SVO
Figure 1: This shows the six primary SVO Slider items
as seen by the subjects.
You receive
85
85
85
85
85
85
85
85
85
Other receives
85
76
68
59
50
41
33
24
15
You receive
85
87
89
91
93
94
96
98
100
Other receives
15
19
24
28
33
37
41
46
50
You
Y
o
1
Other
Ot
You
Y
o
2
Ot
Other
You receive
50
54
59
63
68
72
76
81
85
Other receives
100
98
96
94
93
91
89
87
85
Ot
Other
You receive
50
54
59
63
68
72
76
81
85
You
Y
o
Other receives
100
89
79
68
58
47
36
26
15
You receive
100
94
88
81
75
69
63
56
50
Other receives
50
56
63
69
75
81
88
94
100
You receive
100
98
96
94
93
91
89
87
85
Other receives
50
54
59
63
68
72
76
81
85
You
Y
o
3
4
Ot
Other
You
Y
o
5
Other
Ot
You
Y
o
6
Ot
Other
metric criteria. This new method is referred to as the SVO
Slider Measure. We provide a detailed discussion of this
new measure, along with norming data and evidence of
the new measure’s strong psychometric properties.
2 The SVO Slider Measure
The SVO Slider Measure can be administered as a paper
based choice task or as an online measure. The measure
has six primary items with nine secondary (and optional)
items. All of the items have the same general form. Each
item is a resource allocation choice over a well defined
continuum of joint payoffs. For example, consider a DM
choosing a value x between 50 and 100 inclusive. Her
payoff would be x, whereas the other’s payoff would be
150 − x. The DM would indicate her allocation choice by
marking a line at the point that defines her most preferred
joint distribution (see item 5 in Figure 1, see also Table
7, p. 779). After the DM has marked her most preferred
allocation, she would write the corresponding payoffs resulting from her choice to the right of the item. Although
this step of writing the values is redundant, it serves to
verify that the DM understood the choice task and the resulting allocations.
2.1 Primary SVO slider items
The six primary Slider Measure items are shown in Figure 1. These six items were derived from the six lines
Judgment and Decision Making, Vol. 6, No. 8, December 2011
Figure 2: This figure shows where in the self/other allocation plane the six primary items are from the Slider
Measure.
Altruistic
100
Prosocial
Individualistic
Payoff to other
75
50
25
Competitive
0
0
25
50
Payoff to self
75
100
that fully interconnect the four points corresponding to
the most common idealized social orientations reported
in the literature (altruistic, prosocial, individualistic, and
competitive; see Figure 2). A DM evaluates each of the
items sequentially and for each one indicates her most
preferred joint distribution. The set of responses can then
be scored to yield a single score for the DM, the rank order of her social preferences, and additionally contains a
check for transitivity in her revealed preferences.
The SVO Slider Measure has several advantages. First,
the responses can be evaluated for comprehension (e.g.,
checking the correspondence between the mark on the
distribution line and the written distribution values). Second, the responses can be evaluated for transitivity. Although SVO is a matter of subjective preferences, these
preferences should conform to the elemental requirement
of transitivity. Random responding would likely result in
an intransitive set of responses. Third, the responses yield
a full ranking of preferences over motivations. Fourth, the
measure can be scored in a straight-forward manner to
yield a single index of SVO as follows. The mean allocation for self (Ās ) is computed as is the mean allocation for
the other (Āo ). Then 50 is subtracted from each of these
means in order to “shift” the base of the resulting angle to
the center of the circle (50, 50) rather than having its base
start at the Cartesian origin. Finally, the inverse tangent
of the ratio between these means is computed, resulting
in a single index of a person’s SVO.
(Āo − 50)
(1)
SVO◦ = arctan
(Ās − 50)
Measuring SVO
773
This response format is highly sensitive to individual differences and yields an individual score at the ratio level
of measurement. Assessing SVO in this way also facilitates parameterization and model assessment that is not
possible with other existent measures. Nonetheless, reducing the high resolution score to a nominal category
may be desirable in some cases (e.g., to compare new results to previous studies), and the resulting SVO Slider
angles can be transformed into corresponding categories
with ease as follows.
If a person would choose the option that maximizes the
allocation for the other in each of the six primary items,
the resulting angle would be 61.39◦, indicating perfect altruism. A prosocial DM with inequality aversion would
yield an angle of 37.48◦. A prosocial DM who endeavored to maximize joint gain (and is inequality tolerant)
would yield an angle between 37.09◦ and 52.91◦. The
reason for this range is that this DM would be wholly
indifferent across the entire SVO Slider item that has a
slope of –1 (i.e., the item with endpoints 100, 50 and
50, 100) as it has a constant sum. A perfectly consistent
individualist yields an angle between –7.82◦ and 7.82◦.
The reason for this range is that this particular DM would
be wholly indifferent across the range of outcomes contained in the SVO Slider item that has an undefined slope
(endpoints 85, 85 and 85, 15). A perfectly consistent
competitor yields an angle of –16.26◦.
Given the angles that result from idealized SVO types,
proper boundaries between categories can be derived by
bisecting the respective adjacent ranges. Altruists would
have an angle greater than 57.15◦; prosocials would have
angles between 22.45◦ and 57.15◦; individualists would
have angles between –12.04◦ and 22.45◦; and competitive types would have an angle less than –12.04◦. As it
can be seen, these boundaries are not at intuitive locations. The reason for this is that the Slider Measure only
uses a subset of possible items from the allocation plane
and these items are not symmetrically distributed around
the whole of the ring. Because only an asymmetric set
of items is used here, the resulting convex hull of possible scores is “squished” to the upper-right, relative to
the midpoint of the ring. This characteristic does not adversely affect the validity of the measure.
2.2 Secondary SVO slider items
There are nine secondary SVO Slider Measure items.
This set of items is explicitly designed to disentangle
the prosocial motivations of joint maximization from inequality aversion. The items are defined in the prosocial
area of the self/other allocation plane and have approximately the same magnitude (ranging between 50 and 100
value units) as the six primary items. One noteworthy feature of these secondary items is that all of the distribution
Judgment and Decision Making, Vol. 6, No. 8, December 2011
ranges intersect the diagonal line. This is an important
feature of the set as points on the diagonal line correspond to perfectly equal allocations, i.e., those distributions that minimize inequality between the DM and the
other person. A person motivated to minimize inequality would make allocations on or very near the 45◦ line.
Conversely a person motivated to maximize joint gains
would make allocations at the endpoints, as far from the
diagonal as possible as it turns out, as these allocations
maximize collective earnings. Previous measures of SVO
have not been explicitly designed to make a differentiation between these two motivations. The nine items are
shown in Figures 6 and 7. An example of results from
these items is discussed in Section §3.6.
2.3 Web-based SVO slider measure
In addition to being administrable as a paper-based measure, the Slider Measure has been programmed as an online research tool which can be freely used by any researcher.1 The online measure and supporting material,
as well as the paper based versions of the new measure,
can be found at: http://vlab.ethz.ch/svo/SVO_Slider/.
With the online SVO Slider Measure, items are presented in a random order. Subjects record their choices by
moving a webpage slider input back and forth, changing
the joint allocations until they find their most preferred
joint outcome (see Figure 3 for a screen shot). The online items are dynamic and display information is updated
in real time as the DM moves the slider over the option
space. The choice procedure is the same for all of the
items. After the subjects have participated, the researcher
is sent an email with the datafile attached; the datafile
contains the subjects’ identifying information, date/time
stamp, item order, and all of the DMs’ allocation choices.
3 Psychometric properties of the
SVO Slider Measure
3.1 Slider Measure validation procedure
In order to assess the psychometric properties of the new
SVO Slider Measure it was tested in tandem with the established and most commonly used measures of SVO;
1 Computing results from the SVO Slider (checking for transitivity,
establishing the ranking of preferences, and finding a subject’s SVO
angle) can be somewhat demanding and thus we have developed an
analysis script that automates and simplifies this process. This script
is available for download from the SVO website along with a detailed
tutorial on its use. We also provide an Excel worksheet for researchers
who are interested in quick and basic results.
Measuring SVO
774
Figure 3: Online Slider Measure.
namely the Triple-Dominance Measure (see Van Lange
et al., 1997) and the Ring Measure (Liebrand, 1984).
Fifty-six individuals from various majors were recruited
to participate in a multi-part “decision making study” at
a European university. No deception was used in this research and subjects were guaranteed strict confidentiality for all of their choices. The choices in the experiment were made incentive compatible by means of a
lottery—for each experimental session four subjects were
randomly selected after making their choices and for each
selected person one of their allocation decisions was implemented (i.e., their allocation choice was carried out
such that they received some chosen payoff, as well as
did some other randomly selected person, according to
their actual choice). For all research sessions, subjects
were reminded that their decisions were private and that
there was a real chance that their choices would have a
pecuniary effect upon themselves and some other person
if they happened to be selected by lottery. DMs selected
by lottery were paid privately in cash within a week of
their participation. Each unit of value in the experiment
corresponded to 50 Swiss cents and the average earnings
were 81.70 Swiss francs (US$77) per paid subject.
Three research sessions were run, with one week separating the sessions. Each research session required fewer
than 15 minutes to conduct and used paper based methods. In the first session, subjects completed the 9-item
Triple-Dominance Measure and the 24-item Ring Measure. In the second session, subjects completed the Slider
Measure and the Triple-Dominance Measure. In the third
session, subjects completed the Ring Measure and the
Slider Measure. All of the measures used standardized
values between 0 and 100. This research design allowed us to assess the test-retest reliability of the TripleDominance Measure, the Ring Measure, and the Slider
Measure. It also allowed us to compute the associations between the different measures and establish norming data and convergent validity for the new SVO Slider
Measure.
Judgment and Decision Making, Vol. 6, No. 8, December 2011
Table 1: The percentage of individuals that were assigned
to each of the different SVO categories by the different measurement methods (TD- Triple Dominance, RMRing Measure, SM- Slider Measure), ordered by experimental session.
Prosocial
Individualistic
Competitive
Unclassifiable
Session 1
Session 2
Session 3
TD
59
21
2
18
TD
61
32
3
3
RM
58
36
4
2
RM
53
45
2
0
SM
58
39
3
0
SM
64
34
2
0
Grand
mean
59
35
3
4
Table 2: A cross tabulation showing the frequency of categorization from test-retest between session 1 (S 1) and
session 2 (S 2) for the Triple-Dominance Measure.
S2
Prosocial Individualistic Competitive Unclassifiable
Prosocial
23
1
0
1
Individualistic
2
8
0
0
S1
Competitive
0
1
0
0
Unclassifiable
3
5
1
1
3.2 Results
Table 1 shows the percentage of individuals that were
assigned to each of the different SVO categories by the
different measurement methods, ordered by experimental
session. Across all measurement methods there is a clear
majority type, namely prosocial, occurring about 59% of
the time. Individualist is less common, but found about
35% of the time. Competitive and unclassifiable types
complete the remainder of the sample representing about
3-4% each.
3.3 Reliability
3.3.1 Triple-Dominance Measure test-retest reliability
Measuring SVO
775
Table 3: A cross tabulation showing the frequency of categorization from test-retest between session 1 (S 1) and
session 3 (S 3) for the Ring Measure.
S3
Prosocial Individualistic Competitive Unclassifiable
Prosocial
18
3
0
0
Individualistic
8
12
1
0
S1
Competitive
0
1
0
0
Unclassifiable
0
0
0
0
Table 4: A cross tabulation showing the frequency of categorization from test-retest between session 2 (S 2) and
session 3 (S 3) for the primary SVO Slider items.
S3
Prosocial Individualistic Competitive Unclassifiable
Prosocial
25
1
0
0
Individualistic
3
15
0
0
S2
Competitive
0
1
1
0
Unclassifiable
0
0
0
0
3.3.2 Ring Measure test-retest reliability
Forty-four subjects completed both sessions 1 and 3. Of
those, 30 (18+12+0+0) were categorized into the same
SVO category each time by the Ring Measure, yielding a
consistency of 68%. Further the correlation between the
resulting angles from the test-retest of the ring measure
was r = 0.599.3
3.3.3 Slider Measure test-retest reliability
Forty-six subjects completed both sessions 2 and 3. Of
those, 41 were categorized in the same SVO category
each time by the Slider Measure, yielding a consistency
of 89%. Further the correlation between the resulting angles from the test-retest SVO Slider Measure was r =
0.915.
3.4 Validity
3.4.1 Convergent validity: Categorical agreement
Forty-six subjects completed both sessions 1 and 2. Of
those, 32 (23 + 8 + 0 + 1) were categorized in the
same SVO category each time by the Triple-Dominance
Measure, yielding a consistency of 70% (Goodman and
Kruskal’s gamma.2 = 0.391)
2 As the Triple-Dominance Measure yields a nominal level variable
with more than two categories, a Pearson product moment correlation
coefficient is not an appropriate statistic for assessing its reliability,
hence the non-parametric Gamma statistic is used as an index of testretest association.
Across research sessions, the Triple-Dominance Measure
and the Ring Measure categorized the same subjects into
the same SVO category 67% of the time. The TripleDominance Measure and the Slider Measure categorized
the same subjects in the same SVO category 74% of the
time. The Ring Measure and the Slider Measure catego3 In order to verify the robustness of these results, non-parametric
statistics of association were also conducted in parallel with Pearson’s
r. The non-parametric tests yielded the same pattern of results.
Judgment and Decision Making, Vol. 6, No. 8, December 2011
Table 5: The correlation coefficients between the different sessions and methods. These values show both the
test-retest reliabilities, as well as the cross method correlations (in gray) which address convergent validity.
776
Measuring SVO
Figure 4: The distribution of SVO scores from the Slider
Measure as represented by angles. The dark line is a
smoothed kernel density estimation.
Competitive
Individualistic
Prosocial
Altruistic
0.16
0.14
0.12
Proportion
RM-1
RM-3
SM-2
SM-3
RM-1 RM-3 SM-2 SM-3
1
0.599
1
0.724 0.536 1
0.680 0.641 0.915 1
0.1
0.08
0.06
0.04
rized the same subjects in the same SVO category 75% of
the time.
0.02
0
−25
−15
−5
5
15
25
35
45
Angle from SVO Slider Measure
55
65
3.4.2 Convergent validity: Correlational agreement
The Ring Measure and Slider Measure both produce continuous results (in the form of angles within the self/other
allocation plane), and these results are amenable to computing correlation coefficients across different measures.
Table 5 displays these correlation coefficients, showing both the test-retest reliability of the Ring Measure
(r = 0.599) and Slider Measure (r = 0.915), as well as
the correlations between SVO angles across the different
measurement methods.
The results show that the Slider Measure correlates as
well (if not better) with the Ring Measure as the Ring
Measure does with itself across retests. This is strong
evidence that the methods are measuring the same thing
and further it demonstrates that the Slider Measure is
more reliable than the Ring Measure (the mean correlation between the different methods is r = 0.649 whereas
the test-retest correlation for the Ring Measure is only
r = 0.599).
3.4.3 Predictive validity
In order to evaluate the Slider Measure’s predictive validity, a second study was conducted where different subjects (N = 100) first completed the Slider Measure and
then played a one-shot anonymous Prisoner’s Dilemma
game. Identical to the first study, this study used monetary incentives determined by a lottery. We found a moderate and statistically significant point-biserial correlation
(r = 0.239) between the subjects’ SVO angles and their
choices in the Prisoner’s Dilemma, indicating a positive
relation between SVO angle and cooperation, as would
be expected. These results are consistent in direction and
magnitude with other findings from incentive compatible choice tasks in social dilemmas and measures of SVO
(Balliet et al., 2009).
3.5 Additional results
As noted before, one advantage of the Slider Measure
is its high resolution, as it yields a ratio level of measurement. Previous measures of SVO produce output
as a simple categorization, which is a limitation. Conversely, producing a ratio level variable, the distribution
of observed SVO angles can be plotted and the density of
different orientations can be estimated. Figure 4 shows
this distribution. A LOESS (Cleveland & Devlin, 1988)
smoothed kernel density estimation was made of the distribution of SVO scores to provide some general idea of
its shape. We find a multimodal distribution of SVO types
in our sample. The largest clustering is in the prosocial
region shifted slightly to the left (toward individualistic).
The second clustering is in the individualistic region and
is shifted to the right (toward prosocial). Within this region is the most common SVO score of 7.82◦ which corresponds to perfectly individualistic choices. The density
function trails off to the left, denoting only a few competitive types. As can be seen in the figure, there is substantial variance in the subjects’ SVO angles, beyond what a
nominal level categorization would indicate. Moreover,
the observed variance supports the assertion that a sensitive SVO measure, which produces reliable high resolution data on a continuous SVO scale, is valuable in that it
can capture the rich gradation of social preferences.
As already noted, the transitivity of responses can be
assessed with the Slider Measure. We found that 98% of
our subjects produced completely transitive sets of social
preference choices. This finding stands in stark contrast
to the consistency results from the Ring Measure where
only 55% of the same subjects produced internally consistent results. This would indicate that almost all subjects have well defined social preferences but that the
Judgment and Decision Making, Vol. 6, No. 8, December 2011
Table 6: The full rank orderings of social preferences
from the SVO Slider Measure across sessions. Note that
25% of the decision makers were indifferent between Individualistic and Prosocial allocations when their inferred
preferences are reduced to ranks.
Measuring SVO
777
Figure 5: The distribution of prosocial preferences, ranging from perfect inequality aversion to perfect joint gain
maximization. The most common preference is for joint
gain maximization (29%) but there is substantial variance
in DM’s prosocial preferences.
0.3
Second
preference
Third
preference
Least
preferred
Prosocial
Prosocial
Prosocial
(Individualistic
Individualistic
Individualistic
Competitive
Individualistic
Altruistic
Individualistic
Prosocial)
Competitive
Prosocial
Individualistic
Altruistic
Individualistic
Competitive
Competitive
Prosocial
Altruistic
Prosocial
Competitive
Competitive
Altruistic
Altruistic
Altruistic
Competitive
Altruistic
Percent
27%
25%
13%
25%
4%
2%
4%
Proportion of prosocial subjects
First
preference
0.25
0.2
0.15
0.1
0.05
0
Ring Measure is not particularly well suited to measure
them.
As an additional feature, the full ranking of people’s
social preferences can be obtained from evaluating the six
primary items of the SVO Slider Measure (see Table 6).
These kind of complete ordinal results are not possible
with other common SVO measurement methods. Moreover, information about a DM’s least preferred allocation
may be useful to know when measuring individual differences.
3.6 Separating the prosocial preferences of
inequality aversion and joint maximization
The secondary items from the Slider Measure are designed to differentiate between two different prosocial
motivations: inequality aversion and joint maximization.
As prosocial behavior can arise from both of these underlying motivations, we demonstrate here how to disentangle these motivations using the secondary SVO Slider
items.
In order to identify prosocial DM’s underlying motivations, two mean difference scores were computed for
each prosocial subject from their allocation choices on
the secondary Slider Measure items. The first difference
score was defined as the average normalized distance between the subject’s allocations and the particular allocations that would maximize equality. For example, if a
DM always chose allocations that were on the diagonal
line (see Figure 7), her mean difference score from idealized inequality aversion would be zero, indicating perfect consistency with the preference of inequality aversion. A second difference score was computed for each
subject that was defined as the mean distance between
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Prosocial preferences from inequality aversion (0) to joint gain maximization (1)
her selected allocations and the particular allocations that
maximized joint payoffs for that item. If the mean difference for this second index was zero, it indicates that
the DM’s allocation choices are perfectly consistent with
joint maximization. These values can be meaningfully
aggregated into a single index by computing the ratio of
the first difference score divided by the sum of both difference scores. The result is an index ranging between
0 (indicating allocation choices perfectly consistent with
inequality aversion) and 1 (indicating allocation choices
perfectly consistent with a preference for joint gain maximization).
Results were obtained from the 79 DMs who made
consistently prosocial allocations in the primary and secondary items across both studies. The distribution of individuals’ inequality aversion / joint gain maximization
indices is shown in Figure 5. Several results are noteworthy. First, this distribution suggests that prosocial DMs
are not homogeneous with respect to their more nuanced
prosocial preferences. Some people are striving for maximizing joint gain, whereas others seem to be, at least
somewhat, sensitive to equality between payoffs. Second, while the modal preference is for joint gain maximization, a slim majority of DMs are actually closer to
inequality aversion. This distribution is both non-uniform
and non-skewed with the mean and median at 0.571.
Splitting the sample at 0.5, 54% of DMs would be categorized as inequality averse whereas 45% would be better
described as joint gain maximizers (one person is exactly
at the midpoint of 0.5 and is not categorized). Lastly, the
shape of the distribution suggests that there is greater conformity in how joint maximizing DMs make allocation
choices compared to how inequality averse DMs allocate
resources.
Judgment and Decision Making, Vol. 6, No. 8, December 2011
4 Discussion
Social preferences are of fundamental importance in
understanding interdependent decision making behavior
among people. In order to quantify the degree to which
people care about outcomes for others, it is necessary to
develop reliable measurement methods to assess this construct. Consistent with this goal, we have described a
new measure of SVO and demonstrated that it is quick,
efficient, easy to implement, has very good psychometric properties, yields scores for individuals at the ratio
level, and facilitates comparison to other measures. The
advent of a high resolution measure of SVO opens opportunities for different research streams to use social preferences as a dependent variable. These types of studies
could address questions regarding how context, information, experience, and framing affect peoples’ propensities
to make tradeoffs in resources between themselves and
others. These lines of research could also answer larger
questions like under what conditions is the selfishness axiom a good approximation for explaining human behavior, and when is it insufficient, or even grossly inaccurate.
This new measurement method can serve as a bridge between perspectives informed by Homo economicus and
those perspectives which take descriptive accuracy as a
starting point.
In a broader sense, we would like to encourage scientists interested in human decision making to develop
higher resolution measures. Having more sensitive and
reliable measurement methods is critical for the detection of subtle yet important effects that may result from
changes in context and information. Therefore, we think
that the technique employed with the Slider Measure
could also be useful in the development of related methods for assessing other individual differences, such as risk
perception (e.g., Ganzach, 2000; Ganzach et al., 2008),
or temporal discounting (e.g., Stevenson, 1992). In general, we believe that allowing subjects to explore a range
of well ordered and intuitive options facilitates not only
the revelation of preferences, but also the discovery and
unencumbered expression of those preferences.
References
Au, W. & Kwong, J. (2004). Measurements and effects of
social-value orientation in social dilemmas: A review.
In R. Suleiman, D. Budescu, I. Fischer, & D. Messick,
(Eds.), Contemporary Psychological Research on Social Dilemmas, pp. 71–98. New York: Cambridge University Press.
Balliet, D., Parks, C. & Joireman, J. (2009). Social
value orientation and cooperation in social dilemmas:
A meta-analysis. Group Processes & Intergroup Rela-
Measuring SVO
778
tions, 12, 533–547.
Burkart, J. M., Fehr, E., Efferson, C. & van Schaik, C. P.
(2007). Other-regarding preferences in a non-human
primate: Common marmosets provision food altruistically. Proceedings of the National Academy of Sciences, 104, 19762–19766.
Cleveland, W. S. & Devlin, S. J. (1988). Locally
weighted regression: An approach to regression analysis by local fitting. Journal of the American Statistical
Association, 83, 596–610.
Cohen, J. (1983). The cost of dichotomization. Applied
Psychological Measurement, 7, 249–253.
De Dreu, C. K. W. & Boles, T. L. (1998). Share and
share alike or winner take all?: The influence of social
value orientation upon choice and recall of negotiation
heuristics. Organizational Behavior and Human Decision Processes, 76, 253–276.
Ganzach, Y. (2000). Judging risk and return of financial
assets. Organizational Behavior and Human Decision
Processes, 83, 353–370.
Ganzach, Y., Ellis, S., Pazy, A. & Ricci-Siag, T. (2008).
On the perception and operationalization of risk perception. Judgment and Decision Making, 3, 317–324.
Henrich, J., Boyd, R., Bowles, S., Camerer, C., Fehr, E.,
Gintis, H., McElreath, R., Alvard, M., Barr, A., Ensminger, J., Henrich, N., Hill, K., Gil-White, F., Gurven, M., Marlowe, F., Patton, J. & Tracer, D. (2005).
“Economic man” in cross-cultural perspective: Behavioral experiments in 15 small-scale societies. Behavioral and Brain Sciences, 28, 795–855.
Knight, G. P. & Dubro, A. F. (1984). Cooperative, competitive, and individualistic social values - an individualized regression and clustering approach. Journal of
Personality and Social Psychology, 46, 98–105.
Kuhlman, D. M. & Marshello, A. F. (1975a). Individual
differences in the game motives of own, relative, and
joint gain. Journal of Research in Personality, 9, 240–
251.
Kuhlman, D. M. & Marshello, A. F. (1975b). Individual differences in game motivation as moderators of
preprogrammed strategy effects in prisoner’s dilemma.
Journal of Personality and Social Psychology, 32, 922–
931.
Liebrand, W. (1984). The effect of social motives, communication and group-size on behavior in an n-person
multi-stage mixed-motive game. European Journal of
Social Psychology, 14, 239–264.
McClintock, C. & Van Avermaet, E. (1982). Social values and rules of fairness: A theoretical perspective.
In V. Derlega & J. Grzelak, (Eds.), Cooperation and
Helping Behavior, pp. 43–71. New York: Academic
Press.
Murphy, R. O. & Ackermann, K. A. (2011). A review of
measurement methods for social preferences. Working
Judgment and Decision Making, Vol. 6, No. 8, December 2011
Paper, Chair of Decision Theory and Behavioral Game
Theory, ETH Zürich.
Radzicki, J. (1976). Technique of conjoint measurement
of subjective value of own and other’s gains. Polish
Psychological Bulletin, 7, 179–186.
Roch, S. G., Lane, J. A. S., Samuelson, C. D., Allison,
S. T. & Dent, J. L. (2000). Cognitive load and the
equality heuristic: A two-stage model of resource overconsumption in small groups. Organizational Behavior and Human Decision Processes, 83, 185–212.
Roch, S. G. & Samuelson, C. D. (1997). Effects of environmental uncertainty and social value orientation in
resource dilemmas. Organizational Behavior and Human Decision Processes, 70, 221–235.
Samuelson, C. D. (1993). A multiattribute evaluation
approach to structural change in resource dilemmas.
Organizational Behavior and Human Decision Processes, 55, 298–324.
Stevenson, M. K. (1992). The impact of temporal context
and risk on the judged value of future outcomes. Organizational Behavior and Human Decision Processes,
52, 455–491.
Van Lange, P. A. M., Otten, W., De Bruin, E. & Joireman,
J. (1997). Development of prosocial, individualistic,
and competitive orientations: Theory and preliminary
evidence. Journal of Personality and Social Psychology, 73, 733–746.
Van Lange, P. A. M. & Visser, K. (1999). Locomotion
in social dilemmas: How people adapt to cooperative,
tit-for-tat, and noncooperative partners. Journal of Personality and Social Psychology, 77, 762–773.
Wyer, R. S. (1969). Prediction of behavior in 2-person
games. Journal of Personality and Social Psychology,
13), 222–238.
Zeelenberg, M., Nelissen, R. M. A., Breugelmans, S. M.
& Pieters, R. (2008). On emotion specificity in decision making: Why feeling is for doing. Judgment and
Decision Making, 3, 18–27.
Measuring SVO
779
Table 7: SVO item endpoints and subsequent slopes that
define each of the SVO Slider Measure items.
Item
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Endpoint 1
Self Other
85
85
85
15
50 100
50 100
100 50
100 50
100 50
90 100
100 70
100 70
70 100
50 100
50 100
100 90
90 100
Endpoint 2
Self Other
85
15
100 50
85
85
85
15
50 100
85
85
70 100
100 90
50 100
90 100
100 70
100 90
100 50
70 100
100 50
Descriptive information
Slope
Equation
Undefined
2.33
-0.43
–2.43
–1.00
–2.33
–1.67
–1.00
–0.60
–3.00
–1.00
–0.20
–1.00
–0.33
–5.00
y ∈ [15, 85], x = 85
y = 73 · x − 550
3
y = − 73 · x + 850
7
y = − 17
· x + 1550
7
7
y = −1 · x + 150
y = − 37 · x + 850
3
y = − 35 · x + 650
3
y = −1 · x + 190
y = − 53 · x + 130
y = −3 · x + 370
y = −1 · x + 170
y = − 51 · x + 110
y = −1 · x + 150
y = − 31 · x + 370
3
y = −5 · x + 550
Note: The SVO Slider Measure has embedded in it several items
which are also dictator games. Item number 5 from the primary
set has this structure. From the secondary set, items 8, 11, 13
also have a slope of –1, giving them the same tradeoff rate between the payoff for self and other but over different ranges.
Results from these items can be analyzed separately.
Judgment and Decision Making, Vol. 6, No. 8, December 2011
Figure 6: This shows the nine secondary SVO Slider
items as seen by the subjects.
Figure 7: This figure shows the location of the nine secondary items of the Slider Measure in the self/other allocation plane.
(50, 100)
96
93
89
85
81
78
74
100
Other receives
50
56
63
69
75
81
88
94
100
O
t
Other
You receive
90
91
93
94
95
96
98
99
100
You
Y
o
Other receives
100
99
98
96
95
94
93
91
90
You receive
100
94
88
81
75
69
63
56
50
Other receives
70
74
78
81
85
89
93
96
100
You receive
100
99
98
96
95
94
93
91
90
Other receives
70
74
78
81
85
89
93
96
100
Other
O
t
You receive
70
74
78
81
85
89
93
96
100
You
Y
o
Other receives
100
96
93
89
85
81
78
74
70
You receive
50
56
63
69
75
81
88
94
100
Other receives
100
99
98
96
95
94
93
91
90
You receive
50
56
63
69
75
81
88
94
100
Other receives
100
94
88
81
75
69
63
56
50
You receive
100
96
93
89
85
81
78
74
70
Other receives
90
91
93
94
95
96
98
99
100
You receive
90
91
93
94
95
96
98
99
100
Other receives
100
94
88
81
75
69
63
56
50
(70, 100)
(90, 100)
100
70
You receive
780
Measuring SVO
You
Y
o
7
90
(100, 90)
8
You
Y
o
9
O
t
Other
Payoff to other
O
t
Other
80
70
(100, 70)
You
Y
o
10
60
11
O
t
Other
50
(100, 50)
50
You
Y
o
12
O
t
Other
60
70
80
Payoff to self
90
100
You
Y
o
13
O
t
Other
You
Y
o
14
Other
O
t
You
Y
o
15
O
t
Other
5. It is not recommended, but if for some reason categorical results are preferred to ratio level results,
individual subjects’ scores can be diminished to the
categorical level following this scheme:
•
•
•
•
Altruism: SVO◦ > 57.15◦
Prosociality: 22.45◦ < SVO◦ < 57.15◦
Individualism: –12.04◦ < SVO◦ < 22.45◦
Competitiveness: SVO◦ < –12.04◦
Appendix
Explication of boundary determination
The items from the SVO Slider Measure
SVO angle calculation
The boundaries between categories were derived as
follows:
If a subject would choose the option which maximizes
the other one’s payoff in each of the six primary items, the
resulting angle would be 61.39◦, indicating perfect altruism (see Table 8). Likewise, if a person would choose
the option which maximizes the difference between the
own and the other one’s payoff in each of the six primary
items, the resulting angle would be –16.26◦, indicating
perfect competitiveness (see Table 11). For prosocial subjects, there are two ways in which they could answer the
six primary items perfectly consistent (see Table 9). First,
if a subject would choose the option which minimizes the
difference between payoffs in each of the six items, the
resulting angle would be 37.48◦. Second, if a subject
would choose the option which maximizes joint gain in
each of the six items, the resulting angle would be between 37.09◦ and 52.91◦. The reason for this range is
that this DM would be wholly indifferent across the entire SVO Slider item that has a slope of –1 (i.e., the item
1. Calculate the mean of the payoffs a subject allocated
to herself across the six primary items (Ās ).
2. Calculate the mean of the payoffs a subject allocated to the other person across the six primary items
(Āo ).
3. Subtract 50 from both means: Ās − 50 and Āo − 50.
4. In order to compute the SVO angle, calculate the inverse tangent of the ratio of the mean of the payoffs
allocated to the other minus 50 and the mean of the
payoffs allocated to the self minus 50:
(Āo − 50)
◦
SVO = arctan
(Ās − 50)
Judgment and Decision Making, Vol. 6, No. 8, December 2011
Table 8: Derivation of the SVO angle that would result if
a person would consistently choose the altruistic options
781
Measuring SVO
Table 9: Derivation of the SVO angle that would result if
a person would consistently choose the prosocial options
Endpoint 1 Endpoint 2 Altruistic Choice
Item
Self Other Self Other Self
1
85
85
85
15
85
Other
85
Item
Endpoint 1 Endpoint 2
Prosocial Choice
Self Other Self Other
Self
Other
2
85
15
100
50
100
50
1
85
85
85
15
85
85
3
50
100
85
85
50
100
2
85
15
100
50
100
50
100
3
50
100
85
85
85
85
100
4
50
100
85
15
50
100
4
50
100
85
15
50
5
100
50
50
100
50
6
100
50
85
85
85
85
5
100
50
50
100
100 ↔ 50
50 ↔ 100
70
86.7
6
100
50
85
85
85
85
36.7
Resulting means:
84.2 ↔ 75.8 75.8 ↔ 84.2
Resulting means – 50:
34.2 ↔ 25.8 25.8 ↔ 34.2
Resulting means:
Resulting means - 50:
Resulting angle:
20
61.39◦
37.09◦ ↔ 52.91◦
Resulting angle:
with endpoints 100, 50 and 50, 100) as it has a constant
sum. For the domain of individualism, if a subject would
consistently choose the option which maximizes the own
payoff in each of the six items, this would yield and angle between –7.82◦ and 7.82◦ (see Table 10). The reason
for this range is that this particular DM would be wholly
indifferent across the entire SVO Slider item that has an
undefined slope (endpoints 85, 85 and 85, 15).
The boundaries according to which subjects can be categorized were derived by bisecting the ranges between
the angles that are produced when a subject with one of
the four classical motivational orientations answers the
Slider Measure perfectly consistent. When there is a
range of angles which can be produced by perfectly consistent choice behavior (as is the case for individualistic
and prosocial subjects), the maximum/minimum values
are used for computing the boundaries. Concretely, the
boundaries were calculated as follows:
• Boundary between altruism and prosociality:
61.39◦ + 52.91◦
= 57.15◦
2
Table 10: Derivation of the SVO angle that would result
if a person would consistently choose the individualistic
options
Endpoint 1 Endpoint 2 Individualistic Choice
Item
Self Other Self Other Self
1
85
85
85
15
85
Other
85 ↔ 15
2
85
15
100
50
100
50
3
50
100
85
85
85
85
4
50
100
85
15
85
15
5
100
50
50
100
100
50
6
100
50
85
85
100
50
Resulting means:
92.5
55.8 ↔ 44.2
Resulting means - 50:
42.5
5.8 ↔ –5.8
-7.82◦
Resulting angle:
↔ 7.82◦
Table 11: Derivation of the SVO angle that would result
if a person would consistently choose the competitive options
Endpoint 1 Endpoint 2 Competitive Choice
• Boundary between prosociality and individualism:
37.09◦ + 7.82◦
= 22.45◦
2
• Boundary between individualism and competitiveness:
−7.82◦ + −16.26◦
= −12.04◦
2
Item
Self Other Self Other Self
Other
1
85
85
85
15
85
15
2
85
15
100
50
85
15
3
50
100
85
85
85
85
4
50
100
85
15
85
15
5
100
50
50
100
100
50
6
100
50
85
85
100
50
Resulting means:
90
38.3
Resulting means - 50:
40
–11.7
Resulting angle:
-16.26◦